I Built an AI System That Qualifies Leads, Scores Them, and Books Calls Automatically

Built a Multi-Agent AI Sales Assistant with n8n, PostgreSQL, OpenAI & Cal.com 🚀 Over the past few weeks, I've been building an AI system that can handle the complete lead journey for coaches and consultants: ✅ AI Receptionist → Greets visitors, answers questions, captures name/email, and stores conversation history. ✅ Lead Qualification Agent → Collects information such as business type, challenges, goals, budget, and urgency. ✅ Lead Scoring Agent → Evaluates the lead and assigns a score (Cold, Warm, or Hot) based on qualification data. ✅ Booking Agent → For qualified leads, shares a Cal.com booking link and helps move the conversation toward a discovery call. ✅ Main Workflow Router → Acts as the brain of the system and decides which agent should handle the conversation at each stage. Tech Stack: • n8n • OpenAI GPT-4o-mini • PostgreSQL (memory + CRM) • Supabase • Cal.com One of the biggest challenges was maintaining lead state across multiple conversations while making the experience feel natural instead of like a scripted chatbot. Still working on: - Nurture Agent - Follow-up Agent - Analytics Layer Would love feedback from other n8n builders and automation enthusiasts. What would you add or improve in this architecture? Here is the GitHub repository link - https://github.com/Sceflow-AI/sceflow-ai-sales-system #n8n #automation #aiagents #openai #postgresql #supabase #nocode #buildinpublic